parse fails on non-ASCII Subject

[joee]

This parsing fails with the error "Failed reading: satisfy", I assume because the Subject contains non-ASCII characters.

While this might not be email-RFC-compliant, messages like this do make their way through my postfix -> sieve pipeline and arrive in my inbox. Ideally, I would like a way to process them using a lenient purebred-email parsing option.

import Data.Hex
import Data.ByteString.Char8
import Data.Either
import Data.MIME

main = do
  let e = unhex "5375626A6563743A20C2BD20697320612068616C660D0A546F3A203C6578616D706C65406578616D706C652E636F6D3E0D0A0D0A626F64790D0A" :: Either String ByteString
  let b = fromRight "" e
  let e' = parse (message mime) b
--  Data.ByteString.Char8.putStrLn b
  print e'

[frasertweedale]

There is an RFC (of course) - https://datatracker.ietf.org/doc/html/rfc6532. Implementing it properly would require significant API breakage.

However, for just Subject and other unstructured headers, we can get away without any API change. The header value will be a ByteString that is valid UTF-8. I can get a draft PR for you to review during the weekend.

[joee]

FWIW, I have also noticed this problem with the display name of an email address, such as:

From: non-ASCII here <[email protected]>

I understand that might be more difficult to fix.

[frasertweedale]

@joee please try this PR: #83

Indeed, adding support for internationalised email addresses will be a more intrusive change.

[joee]

This works great. In my little test corpus of 1,000 emails it parses 100%, including UTF-8 in both the subject line and the email display-name.